Methods
In the first phase of this research a search was made for all sandy seashores marked on the Basic Map of Finland, scale of 1:20 000, excluding islands and inland waters. All these shores were visited in summer 1987, together with all the sites described by Lemberg (1933-35) located in present-day Finland. A total of 28 sites (Fig. 3) were then chosen for closer research on the grounds of their evidence of aeolian activity. The history of these coasts was studied by comparing their present state with historical records, especially the above-mentioned data collected by Lemberg, and by examining maps and aerial photographs of different ages.
Fig.
3. Locations of the coastal dune fields studied here (black dots) and their
relations to the sand and gravel formations inf Finland (black), which are mainly
glaciofluvial eskers and end moraines. The Hanko Peninsula is shown separately.
The sites studied are: 1. Hyypiä (Vironlahti), 2. Lappohja (Hanko), 3.
Syndalen (Hanko), 4. Henriksberg (Hanko), 5. Vedagrundet (Hanko), 6. Kolaviken
(Hanko), 7. Tulliniemi (Hanko), 8. Padva (Tenhola), 9. Yyterinsannat (Pori),
10. The Herrainpäivät area (Pori), 11. The Karhuluoto area (Pori),
12. Storsanden (Monäs, Nykarleby), 13. Tisskärssanden (Vexala, Nykarleby),
14. Storsand and Lillsand (Pietarsaari), 15. The Karhi dune filed (Lohtaja),
16. The Vattaja dune field (Lohtaja), 17. The bathing beach of Vattaja (Lohtaja),
18. Hietasärkät (Kalajoki), 19. Letto (Kalajoki), 20. Yrjänä
(Tauvo, Siikajoki), 21. Haikaranhietikko (Tauvo, Siikajoki), 22. Ulkonokanhietikko
(Tauvo, Siikajoki), 23. Koppana (Oulunsalo), 24. Pajuperä (Hailuoto), 25.
Marjaniemi (Hailuoto), 26. Virpiniemi (Haukipudas), 27. Röyttänhieta
(Simo), 28. Tiironhiekka (Simo).
On prograding beaches the ecological succession and the development of the geomorphology can be observed from coastal profiles or transect analyses. A total of 32 vegetation profiles were levelled on the selected dune fields and drawn (Hellemaa 1995). The cover of individual plant species was estimated visually on a 7-point scale (cf. Vartiainen 1980: 16) in one-metre bands from the water line to the forest. The cover values used were (Kalliola 1973: 44) 1 = < 2%, 2 = 2-4 %, 3 = 4-8 %, 4 = 8-16 %, 5 = 16-32 %, 6 = 32-64 %, 7 = > 64 %. Geomorphological observations were made and plant and sand samples were collected from the sites. The development of the morphology and plant cover on the shores of Yyteri and Hanko were observed almost every summer between 1987-1996, and a total of 38 soil samples were taken in summer 1991 along the coast from Tulliniemi on the Hanko Peninsula to Haikaranhietikko on the foreland of Tauvo (Fig. 3). The processes affecting the shores in winter were observed during a field trip to Yyteri on 13 March 1994.
All the samples were classified in the field on geomorphological grounds, and processes observed in the field were also used as a basis for classification. Grain-size properties were studied in 138 sand samples collected from the material under dunes, beach material (in crests of berms), dune ridges, low embryo dunes, incipient foredunes, dune slacks, intermediate dunes, separate hummock dunes, lag material of deflation surfaces, transgressive dunes, older stabilized dunes and cover sand. These samples were dry sieved in the normal manner at successive mesh-size intervals of 1/4 phi unit (1 phi unit = -log2 of the value in mm). Cumulative percentage frequency curves were plotted on log-arithmetic probability graph paper, which made it easy to study deviations from a log-normal distribution and differences between samples.
Grain-size parameters were calculated using the formulae proposed by Folk and Ward (1957). According to Friedman (1961), the grain-size distribution depends on the transporting agent and its amount of energy. The larger the graphic mean grain-size (Mz), the more energy the transporting agent had. Similarly, the graphic standard deviation (Sd) represents the variation in that energy, and the graphic skewness (Sk) shows whether the average energy levels during sedimentation were lower (positive skewness) or higher (negative skewness) than the long-term average. The skewness values for a normal distribution are +0.1 - -0.1. A high graphic kurtosis (KG) value means that the particles in the middle of the grain-size distribution were transported for a longer time than normal. Skewness values for a normal distribution are 0.90 - 1.11 (mesokurtic). Values smaller than this are described as platykurtic, and higher ones as leptokurtic. The validities of differences between groups were tested with Student's t-test, and coefficients of correlation between groups were also calculated. Median (Md) values were obtained from the distribution curves.
Pedogenic processes and their significances were studied using soil samples taken at a depth of 2-5 cm, in order to avoid disturbing factors at the surface (such as ripple marks, litter, lichens and mosses with thick rhizoids). The sand thrown up onto the shore by waves contains large amounts of nutrients, which are transported downwards by the leaching effect of rainwater. It was thus thought that leaching could be most easily detected in samples taken as near the soil surface as possible at different distances from the shoreline. A total of 38 soil samples were analyzed in terms of acidity (pH), amounts of soluble calcium (Ca), phosphorus (P), potassium (K) and manganese (Mg) in mg/l in the laboratory of Viljavuuspalvelu. The samples were not taken from different depths, as the development of a chemically detectable podzol profile takes an average of 200-300 years in the Finnish climate (Jauhiainen 1973: 24) and most of the samples taken were younger than this. The amount of organic material was estimated visually on scale: not detectable, sparse, scattered and abundant. The vegetation of the sampled sites was analysed from plots of one square metre.
The water repellency of the sand surface was measured using the method described by Dekker and Jungerius (1990) in 19 samples collected at Yyteri in summer from a depth of 0-5 cm soon after a rain shower. The water content and the amount of organic material was measured by weighing, drying and ashing parts of the samples. The samples were dried at +60oC for three days and then allowed to adapt to room temperature and air humidity. Water repellency was measured in terms of the water drop penetration time on smoothed surfaces. Three drops of distilled water from a standard medicine dropper were placed on the air-dry surface and their penetration into the soil was timed. The median value (second drop) was used for classification on a scale: wettable, non-water repellent (<5 s), slightly repellent (5-60 s), markedly repellent (60-600 s), severely repellent (600-3600 s) and extremely water repellent (>3600 s).
A total of 148 plots of 1 m2 were distinguished for multivariate analyses of vegetation, representing different coasts and different a priori defined classes. This classification was grounded in the ecological succession and the geomorphology: 1) lower beach, 2) upper beach, 3) berm, covered by aeolian sand, 4) foredune, 5) intermediate dune, 6) dune slack, 7) dry deflation surface, 8) separate hummock dune, 9) damp deflation surface, 10) forest edge, 11) windward slope of a partly forested dune and 12) leeward slope of a partly forested dune. These plots together with the 38 plots from which soil samples were collected were tabulated with the Excel program (these tables are kept in the Department of Geography, University of Helsinki) and these vegetation data were used for further analysis using STATGRAPHICS Plus (1995) multivariate methods.
The vegetation data were first classified by Cluster Analysis. The amount of data was restricted by omitting species that occurred only on one shore and species with a cover always <2 %. After this the number of species was further restricted by Principal Components Analysis. In the case of a dominant indicator species such as Leymus arenarius, about 90 % of the samples were placed in the same class at first, and therefore the furthest neighbour method (using the maximum distance between any two observations within a cluster as the criterion for the minimum distance between clusters) and the Euclidean distance (using the square root of the sum of the squared distances between observations) were selected as clustering methods. Discriminant Analysis was used to compare the whole set of vegetation data with the above a priori classification, and also with the coastal vegetation areas of Finland (based on vascular plants) proposed by Kalliola (1973: 186). Discriminant Analysis determines whether there are statistically significant differences between groups and picks out the independent variables that account for most of these differences.
Certain environmental variables were added to the vegetation data collected with the soil samples, namely northernness (km, in the Finnish coordinate system), distance from the water line (m), height above sea level (m), fluctuation in sea level (cm), rate of land uplift (mm/year), inclination of slope (tangent), proportion of bare sand surface (%, may express the mobility of the sand), tree cover (%), mean grain-size of material (Mz), sorting of material (Sd), soil moisture content (%, occasional rains caused random variation), pH and amounts of soluble calcium, phosphorus, potassium and manganese (mg/l) present. Correlations between these environmental variables were studied by Principal Components Analysis and by calculating correlation coefficients. In addition to these, orientation of the coast (N = 0, W = 15, S = 30, E = 45), exposure (windward slope = 1, flat surface = 2, leeward slope = 3), frequency of trampling and amount of organic material (both on a scale 0-3) were estimated by ranking the sampled sites.
The significance of the environmental variables for the ecological succession was investigated by Canonical Correlations Analysis (STATGRAPHICS 1995), both taking the variables one by one and taking them all together. Logarithmic transformations were used for notably skewed distributions (height, tree cover, Ca and K). The analyses were first performed separately for the tree, shrub and herb layers and for the ground layer of mosses and lichens using all plant species, after which it was focused on the group of the most significant indicator species, as selected by Principal Components Analysis.
The nomenclature mainly follows Hämet-Ahti et al. (1986) for vascular plants, Ahti (1981) for lichens and Koponen (1986, Bryophytina) and Piippo (1987, Hepaticophytina) for mosses. For convenience Empetrum nigrum subsp. hermaphroditum is referred to here as Empetrum hermaphroditum. The 'sand lumps' in the vegetation profiles are formed by fungal hyphae binding the sand grains together. The Cladonia subgenus Cladonia lichens are often listed as one group in the profiles, as their cover is usually very small and there are numerous species. Similarly, the Cladonia subgenus Cladina lichens are sometimes listed as one group, as C. mitis and C. arbuscula are difficult to distinguish, particularly in the field, and can be identified with certainty only chemically, using a PD-colour reagent (Ahti 1981). This chemical analysis was performed only on some of the samples from Yyteri, all of which proved to be C. abuscula. C. mitis is common in the north of Finland, and is found in the south mainly in open places (Ahti 1981: 28), of which coastal dune fields represent one type.